Trump Administration’s AI Policy Shuffle: From Hands-Off to Possible Pre-Release Vetting

The White House has sent contradictory messages about whether new AI systems must undergo vetting. Anthropic’s “Mythos” model, which can uncover vulnerabilities, prompted officials—including NEC director Kevin Hassett—to discuss an executive order that could require Anthropic, OpenAI, Google DeepMind, Microsoft and xAI to obtain clearance before deployment. Industry leaders say they lack clarity on timing, scope and legal basis.

Politico

President Donald Trump began his second term with a light-touch AI stance, but Mythos—capable of patching and exploiting security flaws—prompted a shift to security concerns. Key officials now include chief of staff Susie Wiles, NEC director Kevin Hassett, National Cyber Director Sean Cairncross and Treasury Secretary Scott Bessent. The AI firms involved are Anthropic (CEO Dario Amodei), OpenAI, Google DeepMind, Microsoft and xAI.

The Hill

March Quinnipiac poll shows 80 % of U.S. adults worry about AI’s societal impact. CAISI has evaluated models from OpenAI and Anthropic since 2024 and now has voluntary testing agreements with three labs. Mythos is shared with a few dozen security groups, with similar capabilities expected within a year.

Politico

The White House says any policy will be announced by the president and calls current deliberations “speculation.” Hassett described the possible order as a safety roadmap; Bessent said “maintaining safety” is a core government charge alongside innovation. Wiles pledged to “balance advancing innovation and ensuring security.”

Politico

Lobbyists prefer a voluntary, CAISI-led review over mandatory clearance. The Cato Institute warned an approval system could be an anticompetitive “kill switch.” ITIF’s Castro cautioned uneven approvals could give competitors an advantage in weeks or months.

Politico

Sources differ on whether the administration is moving toward an FDA-style regime or merely exploring voluntary testing. White House officials call the debate “speculation,” while industry insiders cite timelines for an executive order. Legal feasibility and the scope of any vetting remain unclear.

Politico

• “There is no clarity,” — Senior tech lobbyist

Politico

• “Adopting an FDA-style regulatory regime for AI would represent a shocking policy reversal by the Trump administration, and a major about-face on how America has approached software, online speech, and digital commerce.” — Neil Chilson, Abundance Institute

The Hill

• “We’re studying possibly an executive order to give a clear roadmap to everybody about how this is gonna go, and how future AIs that also potentially create vulnerabilities should go through a process so that they’re released into the wild after they’ve been proven safe — just like an FDA drug,” — Kevin Hassett, NEC Director

Politico

• “The White House will continue to lead an America First effort that empowers America’s great innovators, not bureaucracy, to drive safe deployment of powerful technologies while keeping America safe,” — Susie Wiles, White House chief of staff

Politico

Industry expects an executive order in the coming weeks, while CAISI expands voluntary testing. Congressional committees are preparing AI-safety hearings, and officials are urging clearer federal guidance on coordination against AI-enabled cyber threats.

Politico